(507b) Application of Mechanistic Models for the Digital Design and Online Control of Pharmaceutical Processes

Mechanistic models are becoming more commonly applied for Research and Development in the pharmaceutical sector. Traditionally, the output from this activity, namely a validated mechanistic model, which is capable of quantitatively predicting the behaviour of the various Critical Quality Attributes (CQA) for typical batch or continuous pharmaceutical processes for a wide range of Critical Process Parameters (CPP). However, these tools are almost exclusively employed in an offline manner currently, primary aimed at assessing process robustness and variability, with very little subsequent online application of the model for control or soft sensing purposes.

Traditionally, online control companies utilize statistical models that require a significant level of tuning and verification, with the online full-scale plant, using Pseudo-Random Binary Sequences (PRBS) to vary the process parameters and observed the process responses. This requirement for the application of rigorous Model Predictive Control (MPC) using statistical model approaches commonly leads to reduced appetite for uptake of the technology in the Pharmaceutical sector. These online trials for tuning the MPC may result in off-specification productions, potentially leading to significant losses in productivity, as the controllers pushes the process to observe responses at extreme points. However, most of these drawbacks can be overcome by the integration of mechanistic models, developed using laboratory scale data with MPC system, such as that offered by Perceptive Engineering, namely PharmaMV.

In this work we outline, the application of an advanced process modelling tool, namely gPROMS FormulatedProducts, to describe a number of pharmaceutical processes. The process model and the mechanistic model kinetic parameters were validated using process data gathered from the literature and from lab-based experiments. The lab-based process was subsequently used to predict the behaviour of the full scale production scale. In order to make the model predictive, some refinement of the kinetic parameters for secondary nucleation was required using minimal experimental data from the typical plant runs.

The mechanistic model was integrated with PharmaMV to develop and tune the MPC against the mechanistic simulation of the process. The PharmaMV platform was subsequently transferred to the physical process. With this approach the MPC derived from the mechanistic model was utilized to accurately control the defined CQAs, such as final particle attributes, potency or moisture.